ABSTRACT
Low density lipoprotein (LDL) receptor gene is regulated at the transcriptional level by the intracellular level of sterols in animal cells. We have recently identified a 20 bp long region (-145 to -126), designated Footprint 1 (FP1), participating in maximal expression of the human LDL receptor gene in the absence of sterols in HepG2 cells [Mehta, K. D., Chang, R., Underwood, J., Wise, J. and Kumar, A. (1996) J. Biol. Chem., 271, 33616-33622]. To determine the minimal FP1 sequence and to define the critical nucleotides required for function, a series of single nucleotide substitutions were introduced in the FP1 region. Twenty-three independent mutations were analyzed by transfection into HepG2 cells. These studies localize the regulatory region to 14 bp and demonstrate the requirement for essential guanine nucleotides at positions -135 and -136 for FP1 function. Furthermore, transfection studies suggest that the FP1-dependent increase in reporter gene expression is possibly mediated through interaction with the sterol-regulatory element. UV cross-linking and Southwestern blot analysis identified FP1-binding factors of ~50 and 125 kDa, which we have denoted p50 and p125. Mutations of the critical guanine residues (-135/-136) decreased the formation of the specific protein-DNA complex with the FP1 sequence and abolished its binding to the p125. We conclude that direct interaction of the p125 factor with these nucleotides of the FP1 element potentially contributes to FP1-dependent induction of LDL receptor gene expression.
The low density lipoprotein (LDL) receptor plays a central role in the regulation of body cholesterol homeostasis (1 ). LDL receptor is transcribed actively when animal cells require cholesterol and is repressed when sterols accumulate (2 ). Transfection studies have demonstrated that the feedback suppression of the LDL receptor gene by sterols is mediated at the transcriptional level by a 10 bp sequence element in the 5'-flanking region designated sterol regulatory element-1 (SRE-1) (3 ,4 ). The essential elements of this sequence are conserved in evolution as far back as the last common ancestor of humans and frogs (5 ,6 ). Furthermore, mutation of SRE-1 results in constitutively low levels of expression, indicating that this element acts in a positive fashion that activates transcription when intracellular sterol levels are low (7 ,8 ). Detailed mutational analysis of SRE-1 led to the generation of specific probes and identification of SRE-1 binding proteins (SREBP-1 and SREBP-2), members of the basic helix-loop-helix-leucine zipper family of transcription factors (9 ,10 ). Recent evidence indicates that under circumstances of sterol deprivation, cytoplasmic membrane-bound full-length SREBP is proteolytically processed to a soluble N-terminal fragment, containing the transcriptional activation and SRE-1 binding domains, which migrates into the nucleus (11 ,12 ). The transcriptionally active cleaved SREBP binds to SRE-1 and interacts synergistically with adjacent bound Sp1 transcription factor to promote LDL receptor gene transcription in a sterol-dependent manner (13 ). When sterols are abundant, SREBP processing is inhibited, thereby leading to a reduction of transcription (11 ).
Exogenous factors other than ambient sterol levels also play a role in regulating LDL receptor gene expression in vivo. LDL receptor gene transcription is induced by various non-sterol stimuli, including activation of protein kinase C, increase in intracellular calcium, inhibition of protein synthesis, and a variety of cytokines and growth factors in multiple cell types (14 -21 ). In most cases, mitogen-induced increases in LDL receptor gene transcription were observed regardless of exogenous and cellular sterol levels.
Recently, by using a combination of in vivo footprinting and functional assays with human LDL receptor promoter constructs, we have identified a novel cis-acting element, designated FP1, that is required for maximal expression of the LDL receptor gene in response to depletion of sterols (22 ). Analysis of the transcription factor database failed to reveal an identifiable match for the FP1 target sequence. The FP1 site appears to be complex, spanning at least 20 bp and possibly consisting of multiple interacting elements. The current study was initiated to evaluate the effect of single nucleotide substitutions in the FP1 region to identify nucleotides required for FP1-dependent induction, and to use the generated probes to characterize the nuclear factor(s) interacting with this element. The results showed that the central guanine nucleotides (-135/-136) are important for FP1 function and its interaction with the FP1-binding nuclear factors.
[[gamma]-32P]ATP (>5000 Ci/mmol) was obtained from ICN. Polynucleotide kinase was obtained from Ambion, Inc. Enzymes used in plasmid constructions were obtained from New England Biolabs, Boehringer Mannheim, and Life Technologies. Plasmids pGL2 and pSV-[beta]-galactosidase (pSV-[beta]-Gal) were purchased from Promega Inc. Plasmid pGL2 has no defined eukaryotic promoter or enhancer sequences and contains the [beta]-lactamase gene (ampR), the pBR322 origin of replication, and the coding sequence for luciferase. pSV-[beta]-Gal vector was used as a positive control for monitoring the transfection efficiency of HepG2 cells. Standard molecular biology techniques were used (23 ). Fetal bovine lipoprotein deficient serum (LPDS) was purchased from PerImmune Inc. Lipofectamine was purchased from Life Technologies Inc. Dual-light chemiluminescent reporter gene assay system for the combined detection of luciferase and [beta]-galactosidase was purchased from TROPIX, Inc. 25-Hydroxycholesterol and cholesterol were purchased from Sigma and Steraloids, Inc., respectively. All tissue culture supplies were purchased from Life Technologies.
HepG2 cells were routinely grown in RPMI 1640 medium supplemented with 10% fetal calf serum. For transfection experiments, HepG2 cells were seeded at 1 × 106/60 mm dish 1 day in advance. Transfections were performed in triplicate with 0.2 µg of each construct DNA and 0.5 µg of pSV-[beta]-Gal vector and 6 µl of lipopfectamine for 6 h (22 ). Cells were washed and refed with the same medium containing 10% fetal calf serum. Approximately 1 day later, transfected cells were switched to media supplemented with 10% LPDS and incubated for an additional 16-20 h. Finally dishes were washed with phosphate-buffered saline and lysed with 150 µl of luciferase lysis buffer described earlier (22 ). Luciferase and [beta]-galactosidase activities were measured with an automated luminometer (Model 2010, Analytical Luminescence Laboratory) as described earlier (22 ).
To construct single nucleotide substituted FP1 mutants (plasmids 1-23), PCR was performed with synthetic oligonucleotide primers containing HindIII sites (underlined) 5' to the FP1 sequences, using standard PCR conditions. Primers used were as follows: 5'-TACAAGCTTAGAGCTGCACGGGTTAAA-3' (containing the desired nucleotide change) and 5'-TAC
For generation of desired nucleotide substitution within SRE-1 or Sp1, a modified oligonucleotide-directed mutagenesis procedure was performed using the Sculptor in vitro mutagenesis system (Amersham). Oligonucleotides of 20 nucleotides in length containing the indicated nucleotide substitutions were used as a primer on a single-stranded M13 template containing PstI fragment of human LDL receptor promoter. To construct plasmids A-H, HindIII-linkered oligonucleotides were used in the amplification reaction, and the amplified fragments were subcloned in the sense orientation into the HindIII site of pGL2-Basic vector. Combination of oligonucleotides A and B, corresponding to nucleotides -110/-90 and +42/+23 respectively, were used for the construction of plasmids A, C, E and G, whereas combination of oligonucleotides B and C (-145/-123) were employed for the construction of plasmids B, D, F and H.
To obtain the 5'-flanking region of the monkey LDL receptor gene, PCR was carried out on rhesus monkey genomic DNA using two oligonucleotide primers with EcoRI linkers (underlined) whose sequences corresponded to: (i) a portion of the 5'-flanking region of the human LDL receptor gene upstream of the FP1 site (5'-TAC
Cloning of the Xenopus LDL receptor gene by screening a Xenopus genomic library has been reported earlier (5 ). Nucleotide sequence of the region farther upstream of the distal Sp1 site of the Xenopus LDL receptor gene was determined by subcloning the DNA fragment of interest into the PstI site of an M13 vector, followed by sequencing with the dideoxy-chain termination method.
Southwestern blot analysis was performed as described by Singh et al. (24 ), with slight modification. Two hundred micrograms of HepG2 and HeLa nuclear proteins were mixed with sodium dodecyl sulfate (SDS) gel loading buffer and boiled for 3 min. The nuclear proteins were resolved by SDS-polyacrylamide gel. The separated proteins were electrotransfered to a nitrocellulose filter (Schleicher & Schuell) in transfer buffer containing 50 mM Tris-HCl, 40 mM glycine, 0.04% SDS, 20% methanol using a Bio-Rad Trans-Blot Cell at 125 mA at room temperature for 12 h. The nitrocellulose was briefly air-dried, and subsequently, the bound proteins were denatured in 6 M guanidine hydrochloride, 25 mM HEPES, pH 7.6, 60 mM KCl, 1 mM EDTA, and 0.5 mM dithiothreitol (DTT) for 10 min at 4oC with gentle shaking. The proteins were renatured by serial step-wise dilutions of the guanidine hydrochloride at 4oC. Once the guanidine hydrochloride was completely diluted out of the HEPES buffer, the nitrocellulose was incubated in BLOTTO buffer [5% Carnation nonfat dry milk, 0.5 µg/ml sonicated calf thymus DNA, 1 mM DTT, 25 mM HEPES, pH 7.6, 60 mM KCl, 1 mM EDTA] at room temperature for 1 h with gentle shaking. The filter was washed twice with 0.5% Carnation milk BLOTTO buffer for 10 min at room temperature and was probed in the same buffer with 1 × 106 c.p.m./ml radiolabeled FP1A/FP1B or FP1Am/FP1Bm (nucleotide sequences described below). The filter was washed three times in 200 ml of 25 mM HEPES, pH 7.6, 60 mM KCl, 1 mM EDTA, and 1 mM DTT for 10 min each wash, blotted dry, and subjected to autoradiography for 16 h at -70oC with an intensifying screen to visualize the protein-DNA interactions.
The photoaffinity probes used in UV cross-linking experiments were 32P-labeled 27 bp double-stranded modified wild-type oligonucleotides FP1A/FP1B (-148 to -121) or modified mutant oligonucleotides, FP1Am/FP1Bm (mutations indicated by small letter in the sequence given below) which were synthesized with bromodeoxyuridine residues (U = 5-bromodeoxyuridine) encompassing the FP1 region of the human LDL receptor promoter (BIOSYNTHESIS, Inc.). The nucleotide sequences of the wild-type and the mutant oligonucleotide pairs in the FP1 core sequence are:FP1A/FP1B5'-TCAGAGCTTCACGGGTTAAAAAGCCGA-3'3'-AGTCTCGAAGUGCCCAAUUUTTCGGCT-5'FP1Am/FP1Bm5'-TCAGAGCTTCACttGTTAAAAAGCCGA-3'3'-AGTCTCGAAGUGaaCAAUUUAACGGCT-5'
The binding reactions were performed as described earlier (22 ). After irradiation with a UV lamp (254 nm) for 5 min, the samples were loaded onto an 8% SDS-polyacrylamide gel, and run together with molecular weight markers. Radioactively labeled protein-DNA complexes were identified by subsequent autoradiography. For competition studies, an unlabeled unmodified oligo pair (FP1A/FP1B) was added to the binding reaction.
Double-stranded bromodeoxyuridine modified FP1A/FP1B or FP1Am/FP1Bm oligonucleotides used in the UV cross-linking experiments were end-labeled using [[gamma]-32P]ATP (22 ). Nuclear proteins (5 µg) were incubated with the labeled probes for 30 min at room temperature in reaction buffer (20 mM HEPES, pH 7.6, 75 mM KCl, 0.2 mM EDTA, 20% glycerol) and 1 µg poly(dI-dC)-poly(dI-dC) (Pharmacia) as non-specific competitor. Protein-DNA complexes were resolved on 5% non-denaturing polyacrylamide gels and visualized by autoradiography.
Conservation of the essential nucleotides during evolution provides a general short-cut method to identify important nucleotides in a given regulatory element of a promoter (5 ). To examine the importance of individual nucleotides within the FP1 region on the basis of sequence conservation, the LDL receptor promoter sequences of two additional species were determined. The monkey and Xenopus LDL receptor promoter sequences are shown in Figure 1 A and B, respectively. The monkey LDL receptor promoter contains SRE-1, Sp1 and FP1 sites virtually identical to their human counterparts. Conservation of the entire FP1 sequence (Fig. 1 C) and its location relative to SRE-1 (not shown) among mammals lend further support for its role in LDL receptor gene regulation. Interestingly, the Xenopus LDL receptor promoter contains all essential nucleotides of repeats 1-3 that are required for the binding of transcription factors SREBP and Sp1 (5 ), although there appears to be no sequence at the expected position or neighboring region that is homologous to the FP1 region. Conservation of these three repeats in all the species examined indicates that the basic mechanism(s) regulating LDL receptor gene transcription by sterols is conserved between amphibian and mammal, but the mammalian receptor gene may have acquired an additional regulatory element (22 ), suggesting a more complex regulatory mechanism.
To understand the mechanism of enhancement of LDL receptor gene transcription by FP1 element, the effect of non-functional SRE-1 and proximal Sp1 site was investigated on FP1 function. Reporter plasmids carrying a single nucleotide substitution either in SRE-1 or Sp1 were constructed and tested for FP1-dependent increase in transcription in response to depletion of sterols in transfected HepG2 cells. As shown in Figure 2 , the presence of the FP1 site resulted in an ~3-fold induction of reporter gene expression (compare plasmid A with B). Similar increase was also observed in plasmid D containing mutated proximal Sp1 site (compare plasmid C with D), even though inactivation of this site had resulted in a dramatic decrease in reporter gene expression. Most importantly, FP1-dependent increase was not observed in plasmid F containing a specific nucleotide substitution that inactivates SRE-1 function (compare plasmid E with F). In contrast, a nucleotide substitution that had no effect on its function did not change FP1-dependent increase of reporter gene expression (compare plasmid G with H).
In view of high conservation of the FP1 sequence in mammalian LDL receptor promoters, essential nucleotides within the FP1 region were determined by measuring the effect of nucleotide substitutions on the functionality of the FP1 site in HepG2 cells. We constructed a series of plasmids in which a single nucleotide within the FP1 region was altered by a transversion (from purine to pyrimidine or vice versa) so that the native spacing of the LDL receptor cis-acting elements is retained. All of these substitutions were made within the sequence extending from -146 to -124, which includes the FP1 site. PCR products containing the desired mutations were incorporated into plasmid pGL2, and assayed for promoter activity by transfection into HepG2 cells. The wild-type sequence and the corresponding transversion mutations introduced at each position are shown at the bottom of Figure 3 . All mutant promoters were independently transfected into HepG2 cells along with the control pSV-[beta]-Gal plasmid. Pools of transfected cells were incubated for 24 h in the absence of sterols prior to enzymatic assays. The ratio of [beta]-galactosidase-normalized luciferase activity to wild-type FP1 activity was calculated for each of the substitution mutation and is plotted in histogram form in Figure 3 , where the ordinate represents relative transcription and the abscissa represents relative DNA sequences from the human FP1 site. Consistent with previous results (22 ), the presence of the FP1 site resulted in ~3-4-fold induction of reporter gene expression (compare plasmids A and B).
Based on the effect of a nucleotide substitution on FP1 function, mutants could be placed into three classes. The first class of mutants showed a major disruption of the FP1-dependent induction and gave rise to luciferase levels that were approximately the same as those seen when the entire FP1 sequence was either scrambled or deleted (22 ). This class of mutation includes mutants 11 and 12. It is significant to note that transversion mutations at nucleotides -135 and -136 abolished the FP1-dependent induction. The second class of mutations mildly reduced (20-50%) luciferase levels as compared with construct A, indicating that this sequence did not make a major contribution to the FP1 function. Examples include mutants 9, 13, 15, 20, and 22. In the third class, transversion mutations had no effect. For example, eight nucleotides (-146 to -139; mutants 1-8) within the FP1 site were apparently not important for FP1 function since transversion mutations had no effect on FP1 function. A similar lack of effect on nucleotide substitution was observed for mutants 10, 14, 16, 17-19, 21 and 23.
To identify the nature of the nuclear factor(s) interacting with the crucial guanine residues, a bromodeoxyuridine-containing mutant oligonucleotide pair, FP1Am/FP1Bm, was created by changing both guanine residues (-135 and -136) to thymines, and the modified labeled probe was used as a probe in the UV cross-linking reaction. As shown in Figure 5 A, the p125 factor showed loss of binding with the modified mutant probe (complex A') without significant reduction in the binding of p50 (complex B') in comparison with the wild-type modified FP1A/FP1B probe. At the same time, EMSA analysis of the same probes using HepG2 nuclear extracts was done to test the effect of this particular mutation on the formation of FP1 sequence-specific protein-DNA complex. In agreement with the UV cross-linking studies, the FP1Am/FP1Bm mutant oligonucleotide pair showed reduction in the formation of complex I without affecting the formation of complexes II and III (Fig. 5 B). Thus, introduction of nucleotide substitutions of the central guanine residues interrupted binding to the p125 factor and decreased reporter gene expression in transient transfection assays. These results strongly implicate these guanine residues as part of the recognition site for the binding of the p125 nuclear factor.
In the present study, we have defined the critical nucleotides that are required for FP1 function in the context of the intact LDL receptor promoter and characterized the DNA-binding proteins complexing with this element. Results presented demonstrate that the nucleotides -138 to -125 of the FP1 region carry most of the infomation necessary for its function in HepG2 cells. Though the entire FP1 sequence is highly conserved in mammals, the crucial region located at nucleotides -135 and -136 is represented by mutants 11 and 12. Nucleotide substitutions at these positions resulted in significant loss of FP1-dependent induction suggesting that the FP1 site acts as a binding site for a conditionally positive activator protein. This interpretation agrees well with the conclusion drawn from in vivo footprinting studies (22 ). Furthermore, lack of FP1-like sequence in the Xenopus LDL receptor promoter raises an interesting possibility that this element is relatively new in comparison with the Sp1 and SRE-1 sites and may have been added to the mammalian receptor genes during the last 350 million years.
No significant similarities with other transcription factor binding sites were identified for FP1 sequence from a computer-assisted homology search in GenBank and other data bases. However, the FP1 element contains a CACGGG sequence (-139 to -134) that is also the core half (
Our results also have begun to address the mechanism of FP1-dependent increase in LDL receptor gene transcription in response to depletion of sterols in HepG2 cells. FP1 can affect transcriptional activity directly or indirectly. In response to depletion of sterols, direct interaction of the FP1-binding factor(s) with the transcriptional machinery can enhance transcription. Alternatively, FP1 can modulate transcription indirectly through interaction with the downstream SRE-1 and/or Sp1 sites. The results presented here are consistent with this latter mechanism and strongly suggest that the enhancement of the LDL receptor gene transcription in response to depletion of sterols by FP1 may require SRE-1. Transfection studies revealed that a single nucleotide substitution that specifically inactivated SRE-1 also abolished the FP1-dependent increase in transcription. In contrast, plasmids carrying either an inactive proximal Sp1 site or a nucleotide substitution that does not affect SRE-1 function exhibited FP1-dependent increase in reporter gene transcription. The observed interaction between the FP1 and SRE-1 sites would be consistent with the increased in vivo protection of both the sites from dimethylsulfate attack in response to depletion of sterols (22 ). These results also support the previous studies suggesting a central role for the SRE-1 in influencing the binding of other transcription factors to human LDL receptor promoter in response to depletion of sterols (31 -33 ).
An emerging theme in regulation of gene expression is that transcriptional control of gene expression can result from physical and functional interactions between transcription factors. Events occurring at the LDL receptor promoter are unknown and presently under intense investigation (reviewed in ref. 33 ). The present study has generated probes that can be used to purify the physiologically relevant FP1-binding factor(s), and thus to investigate interactions with the transcription factors SREBP and Sp1 that bind to the neighboring sequences.
This work was supported by a research grant from theNational Institutes of Health (HL51592-04). We wish to thank Dorothy Iwanski for sequencing the Xenopus promoter and Dr Amit Kumar for his help in Southwestern blotting. We also thank Drs Randy Haun and Bob Reiss for critical review of the manuscript. We acknowledge Dr Patty Wight for use of the automated luminometer.
REFERENCES